Prediction of Air Quality using Type-1 and Type-2 Fuzzy Systems

Abstract

In India, air pollution is a major environmental concern with significant implications for public health and environmental sustainability. This research presents in-depth analysis of Air Quality Index (AQI) prediction for Delhi, engaging a fuzzy logic approach with Type-1 Fuzzy Sets (T1FS), General Type-2 Fuzzy Sets (GT2FS), and Interval Type-2 Fuzzy Sets (IT2FS). The study aims to estimate the correctness and handling fuzzy logic models in capturing the complex and uncertain nature of air quality parameters. The research begins with the collection and preprocessing of historical data related to key air pollutants. Linguistic variables are defined for each input parameter and the output (AQI) using membership functions in T1FS, GT2FS, and IT2FS. Fuzzy rule bases are generated, incorporating expert knowledge and data-driven insights, forming the foundation for the fuzzy inference systems. Training and testing phases are performed using distinct datasets, allowing for a challenging evaluation of the predictive performance of each fuzzy system. Performance metrics, including Root Mean Squared Error (RMSE) and Mean Absolute Error (MAE), are engaged for a quantitative comparison. The findings not only highlight the strengths and weaknesses of T1FS, GT2FS, and IT2FS but also offer guidance for the aftermath of on-target and reliable air quality prediction systems. Additionally, this research not only advances the understanding of fuzzy logic applications in environmental science but also provides practical significance for policymakers and stakeholders working towards air quality management in urban areas.